Dynamic prediction of effort reallocation in mixed fisheries

A discrete choice model is applied to determine how fishing effort is allocated spatially and temporally by the English and Welsh North Sea beam trawl fleet. Individual vessels can fish in five distinct areas, and the utility of fishing in an area depends on expected revenue measured as previous success (value per unit effort) and experience (past fishing effort allocation), as well as perceived costs (measured as distance to landing port weighted by fuel price). The model predicts fisher location choice, and the predictions are evaluated using iterative partial cross validation by fitting the model over a series of separate time-periods (nine separate time-periods). Results show the relative importance of the different drivers that change over time. They indicate that there are three main drivers throughout the study, past annual effort, past monthly effort in the year of fishing, and fuel price, largely reflecting the fact that previous practices where success was gained are learned (i.e. experience) and become habitual, and that seasonal variations also dominate behaviour in terms of the strong monthly trends and variable costs. In order to provide an indication of the model’s predictive capabilities, a simulated closure of one of the study areas was undertaken (an area that mapped reasonably well with the North Sea cod 2001 partial closure of the North Sea for 10 weeks of that year). The predicted reallocation of effort was compared against realized/observed reallocation of effort, and there was good correlation at the trip level, with a maximum 10% misallocation of predicted effort for that year.